Ind. Eng. Chem. Res.

1997, 36, 771-783 771

Fundamental Liquid Flow Correlations for the Computation of

Design Parameters for Ordered Packings

Sanat Shetty and Ramon L. Cerro*

Chemical Engineering Department, The University of Tulsa, Tulsa, Oklahoma 74104-3189

The foundations have been laid out for estimating design parameters in ordered packing
materials based on fluid mechanics and mass transfer without any adjustable parameters.
Design parameters for heat- and mass-transfer processes over ordered packing materials can
be estimated using the knowledge developed on the study of flow patterns and velocity parameters
for thin viscous films over complex surfaces. Velocity profiles can be integrated to calculate
liquid holdup and interfacial areas. The effect of the size of the structured packing as well as
liquid properties on these parameters was explored. Flow visualization experiments were
performed to determine the path followed by tracer particles in the liquid film flowing over the
inclined corrugations of a packing element. Contact times for tracer particles to travel one wave
of the packing element can be evaluated using the packing geometry and the path followed by
the particles. Mass-transfer coefficients for the liquid phase were estimated using a penetration
theory model. These parameters were used to predict height of a transfer unit (HTU) and height
equivalent to a theoretical plate (HETP) for typical commercial packing materials and compared
with existing experimental data. The results shown here conform to a basic model of flow and
mass transfer in ordered packings that can be used for research and design purposes.

1. Introduction rostructure in the way of channel geometry generated

Despite its inherently low thermodynamic efficiency, by corrugations configures the mixing cells for the bulk
distillation is a unit operation that has been used for a flow of the gas phase. The microstructure is present
long time and continues to be the primary method of due to the surface features of the packings and is on
separation in processing plants. The preeminence of the order of 0.003-0.008 m (0.1-0.25 in.). For the wire-
distillation for separation of fluid mixtures is not mesh packings the meshing provides the microstructure.
accidental but fundamental in nature and is therefore In corrugated sheet packings the surface is usually
unlikely to be displaced in the near future (Kister, 1992). grooved, lanced, or textured to provide the microstruc-
According to Fair and Bravo (1990), one of the most ture. The microstructure plays an important role in
notable phenomena observed by distillation column sustaining a stable wetting liquid film on the packing
designers over the last 2 decades has been the move- surface. Some versions of commercial packings, regard-
ment of the packed column into direct competition with less of whether they are made of corrugated sheets or
the cross-flow tray column. Structured packings along wire mesh, have regular perforations on the surface. In
with random packings and ordered grids are the three some cases the surfaces may also be chemically treated
primary packing types used in packed columns. As the to promote lower contact angles and improve wetting.
performance characteristics of these structured or or- While most of the packings have a triangular cross-
dered packings is better understood, their popularity sectional area, some packings like the Montz-B1 have
has risen from just niche applications to the point where a rounded top. The reader is referred to the review of
they are one of the most popular column internals in Kister (1992) for an analysis of the features of different
use today. These packings have now become very commercial structured packings available today.
competitive with conventional column internals, espe- State-of-the-art design of packed columns involves
cially for revamps. They have a well-defined structure geometric data, physicochemical and rate data, adjust-
and outperform random packings in difficult separations able parameters such as temperatures and pressures,
with pressure gradient constraints, low relative volatili- and a set of design parameters such as specific inter-
ties, and/or small liquid holdup. Sometimes, the distil- facial areas, heat- and mass-transfer coefficients, pres-
lation capacity can be increased by as much as 30-50% sure drop, and liquid holdup. Design parameters cannot
with no adverse effects on separation efficiency simply be easily manipulated by varying adjustable parameters
by removing the trays and installing structured pack- because of the complex manner in which they are
ings (Kurtz et al., 1991). interrelated. The primary objective of this work is to
Ordered packings are usually made of corrugated lay the foundations for the prediction of design param-
sheets that may have been mechanically or chemically eters on the basis of flow patterns and transport
treated to improve their wetting characteristics. Or- parameters. Unlike empirical correlations, this ap-
dered packings have a very well-defined mechanical proach results in a better understanding of the role of
structure which includes both a macrostructure and design parameters and their interaction.
microstructure. The macrostructure is due to the cor- Description of Previous Work. Models describing
rugations of the packings and has a characteristic the performance of packed columns that use ordered
dimension from 0.01 to 0.05 m (0.5-2 in.). The mac- packings have been developed and implemented with
some measure of success. The main variables of interest
* To whom correspondence should be addressed. Tele- are the pressure drop, maximum flow capacity, and
phone: (918) 631-2978. Fax: (918) 631-3268. E-mail: RLC@ mass-transfer efficiency. Pressure drop is often used
[Link]. to specify packed tower capacity; however, interpreta-
S0888-5885(96)00627-6 CCC: $14.00 © 1997 American Chemical Society
772 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

tion and application of this criterion are not easy. For

decades, generalized pressure drop correlation charts
(GPDC) have been the industry standard for predicting
pressure drop. The GPDC abscissa is the flow param-
eter, which is the ratio of liquid to vapor kinetic energy,
while the ordinate parameter describes the balance
between the vapor momentum force that acts to carry
liquid droplets up and the gravity force. Kister (1992)
made an extensive review of available GPDC for differ-
ent packing types. In addition to empirical correlation
or interpolation of experimental data, two theoretical
approaches have also been used. The channel model,
which attributes pressure drop to the resistance to
liquid flow in the many parallel channels in the packed
bed, has been used for structured packings by Bravo et
al. (1986). The particle model, which attributes pres-
sure drop to friction losses, has been used by Stichlmair
et al. (1989) for structured packings.
The maximum operational capacity (MOC) is defined
as the maximum vapor rate that provides normal
efficiency of a packing. The MOC is a useful upper-
capacity criterion which is related to packing efficiency.
However, generating MOC data requires expensive Figure 1. Sketch of solid surface and definition of geometric
variables.
efficiency measurements, and MOC data are conse-
quently not widely available. In addition to correlations
available from manufacturers, there is at least one coefficients are different from the actual system of
published empirical correlation by Kister and Gill (1991) interest. More recently, Stasiek et al. (1996) conducted
that is used for both random and structured packings. an experimental and numerical study of flow and heat
transfer in a corrugated geometry under transitional
Because the mass-transfer performance of the packed
and weakly turbulent conditions. They observed the
column is heavily dependent upon the liquid distribution
best overall agreement with measured friction factors
in the packed bed, there have been very few attempts
and local heat transfer coefficients using either a low
to model mass-transfer performance based on fluid flow
Reynolds number k- model or large eddy simulations.
patterns. Packing mass-transfer efficiency is character-
Understanding flow patterns is the first step for
ized by the height equivalent to a theoretical plate
formulating mass- and energy-transfer models. In
(HETP). HETP predictions are commonly based on
addition, operating variables such as liquid holdup and
mass-transfer models, rules of thumb, or data interpola-
interfacial area, are needed for the development of mass-
tion. A mechanistic model for structured packings was
and energy-transfer models. The only obstacle left on
published by Bravo et al. (1985). The model was based
the design of separation systems after the fluid mechan-
on a wetted-wall analogy and focused on gauze-type
ics have been solved is the ability to solve the associated
structured packings. Effective liquid and gas velocities
mass- and energy-transfer problems.
were defined using packing geometry, and Higbie’s
theory was used to predict the liquid-side mass-transfer
coefficient on the assumption that the packing surface 2. Theory
was fully wetted. Later this model was extended (Fair Zhao and Cerro (1992) and Shetty and Cerro (1993)
and Bravo, 1990), introducing an empirical factor to characterized the free surface position of a liquid film
account for the wet fraction of the geometric area. flowing over model test surfaces which simulate the
Gaiser and Kottke (1989) analyzed local convective macro- and microstructure of ordered packings. The
heat and mass transfer in corrugated passages. Nawroc- solid surfaces used in their experiments were made of
ki et al. (1991) proposed a set of models to describe liquid alternating half-circles, triangles, and stacks of cylindri-
distribution and, consequently, mass-transfer perfor- cal rods. The amplitude of the half circles on the solid
mance of ordered packings. Their model was used to surface (Figure 1) was equal to their radius A ) 6.35 ×
investigate the effect of packing height, liquid load, 10-3 m (0.25 in.) for the macrostructure and A ) 1.588
initial maldistribution, and solid-liquid contact angle × 10-3 m (0.0625 in.) for the microstructure. The
on the packed column performance. Wetted area and respective wavelengths were a ) 0.0254 m (1 in.) and a
mass-transfer coefficient were the primary predicted ) 6.35 × 10-3 m (0.25 in.). Data obtained in this work
variables, and their computations agree well with fall in the range 0.1 < δ < 0.4, where δ ) h*/A is the
literature values. Hanley et al. (1994) have proposed a ratio of the Nusselt film thickness to the solid surface
model based on an analogy to electrical percolation on amplitude. These data span the entire range of δ found
a conductor/insulator lattice. The model has been used in industrial practice for flow over macro- and micro-
to predict mass transfer coefficients, dynamic liquid structures. A number of experiments were done with
holdup, and interfacial areas. There are few investiga- the corrugations of the macrostructure horizontally
tions reporting on the actual measurement of interfacial aligned, while other experiments were done with cor-
areas in structured packings. The paper by de Brito et rugations inclined at 45o with respect to the horizontal.
al. (1994) used CO2 absorption into NaOH solutions to For all experiments, the value of the wavenumber (ø )
calculate/measure interfacial areas. Hanley et al. (1994), A/a) was constant and equal to ø ) 0.25.
however, warned against using experimental methods Results shown in this subsection are used as an
of determining interfacial areas in which the relative introduction and mainly for comparison purposes. De-
magnitudes of the liquid and vapor phase mass-transfer tails on the development of equations and solutions can
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 773
be found in Shetty (1995). For the solid surface kept very small or the Reynolds number can be relatively
with its corrugations horizontal, a two-dimensional large, in a first approximation capillary or inertial
stream function, Ψ(x,y), based on a single unknown contributions cannot be neglected. The zero-order
function, h(x), the local film thickness, is defined as perturbation is a third-order nonlinear ordinary dif-
(Shetty and Cerro, 1993): ferential equation involving the dimensionless film

[ ]
thickness and capillary and periodic body forces. The
2
3q (y - ys) (y - ys)3 leading number 1 is the largest viscous term of the
Ψ(x,y) ) (1) linear momentum balance in the direction of motion.
2 h(x)2
-
3h(x)3 The term that has the Capillary number and the
derivative of the Gaussian curvature in it is the capil-
where q is the liquid flow rate per unit packing width. lary contribution to the pressure field. The terms with
This streamline function can be used to compute the NRe included are the inertial terms. Except for the
physical components of the velocity vector: singularity at x ) xc, the curvature of the solid surfaces

[ ]
used in this work was constant and their derivative is
∂Ψ 3q y - ys (y - ys)2 zero.
Vx ) 2 (2)
The Capillary number and the thickness parameter
) -
∂y 2h(x) h(x) h(x)2
δ are related through the definition of the Bond number,

[ ]
2
(y - ys)3 NBo ) (FgA2)/δ ) (3NCa)/δ2. For the macrostructure
∂Ψ 3q dh (y - ys) surface used in this work, the values of the Bond
Vy ) - 2 (3)
2 dx
) -
∂y h(x)3 h(x)4 numbers are NBo . 1 and the capillary term can be
safely neglected. For small film thickness and large
Equations 2 and 3 satisfy continuity, the nonslip solid surface amplitude and wavelength, as was the case
condition at the solid surface, and a simple version of for flow over the macrostructure, the film evolution
the free-shear boundary condition at the free surface. equation reduces to the following algebraic expression:
For a Cartesian coordinate system defined as shown in
Figure 1, the position of the solid surface is fully 1 - φ03 cos θ ) 0 (6)
described by the following function:
where
ys ) yc ( xR2 - (x - xc)2 (4)

Equation 4 describes alternating half-circles centered cos θ )

xR2 - (xs - xc)2
at x ) xc and y ) yc, respectively. A dimensionless film R
evolution equation is obtained by substitution of eqs 2 The velocity profile for the flow of a viscous film over
and 3 into the x- and y-components of the Navier- the macrostructure of a periodic surface, such as the
Stokes equations (Shetty, 1995). Dimensionless vari- surfaces used in this work, can therefore be approxi-
ables and parameters are mated by a sequence of quasi-steady-state Nusselt
solutions for viscous film flow over a surface of continu-
h(x) ys(x)
(3νqg)
1/3
φ(x) ) h* ) Ys(x) ) ously varying inclinations. The dimensionless film
h* h* thickness is given by

a2
[x R
]
1/3
x y φ0(x) ) (7)
X) Y) H )H
a h* h* 2
R - (xs - xc) 2

Attempting a numerical solution to the complete form When combined with eqs 2 and 3, eq 7 allows a simple
of the film evolution equation would be a formidable but remarkably accurate representation of the velocity
task. A perturbation solution was developed assuming field over a periodic surface.
that the perturbation parameter is equal to the product Comparison of experimental and theoretical film
of the thickness parameter and the wavenumber, ú ) thickness profiles was done using the experimental film
δø, which in this case is a small number, 0.025 < ú < thickness normal to the solid surface at every point.
0.1. Rearrangement of the film evolution equation in Since the test surface is made of a collection of half-
powers of the perturbation parameter leads to a singular circle surfaces, the normals to the surface pass through
perturbation: the center of the circle. Measured experimental film
thickness is transformed into normal film thickness
1 - φ3 cos θ - using a computer algorithm based on a geometrical

( )
NReφ′ construction (Shetty, 1995). Figure 2 shows free surface
φ4θ′ cos θ
ú + φ3φ′ sin θ + + φ3Ys sin θ - profiles for flow over the macro-surface for a wide range
10 2 of flow rates ranging from 5.25 × 10-6 to 3.55 × 10-5

( )
m3/(s‚m). The free surface profiles in Figure 2 are
φ3 shown using a laboratory coordinate system, and, as a
ú3 (φ′′′ + Ys′′′) ) 0 (5)
3NCa consequence, it is difficult to appreciate the changes in
film thickness for different flow rates. The normal film
In the following subsections we will introduce results thickness in Figure 3, shown on a curved coordinate
derived from eq 5 for the macro- and microstructure of normal to the solid surface, is distinctly doubly periodic;
packings with corrugations normal to the directions of i.e., two maximums in film thickness are observed over
flow and for packings with corrugations inclined with one period of the solid surface. Figure 3 compares the
respect to the direction of flow. film thickness calculated using the small NRe perturba-
Flow Patterns over the Macrostructure of Or- tion solution and the experimental film thickness nor-
dered Packings. Since the Capillary number can be mal to the solid surface. The agreement between
774 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

Figure 2. Free surface profiles for flow over the macrosurface.

δ: O, 0.108; small solid box, 0.112; 9, 0.1248; [, 0.1336, *, 0.1392, Figure 5. Free surface profiles for flow over the microstructure.
g, 0.146; 0, 0.148; ], 0.1572; +, 0.1608; ×, 0.1624; ‚‚‚, 0.1652; δ: O, 0.0215; small solid box, 0.0209; 9, 0.0267, [, 0.0347; *,
- ‚ -, 0.168; s, solid surface. 0.0447; g, 0.0575; 0, 0.0695; ], 0.071; +, 0.0829; ×, 0.1088.

Figure 3. Comparison of experimental normal film thickness with Figure 6. Comparison between experimental and predicted film
the perturbation solution for flow over the macrosurface: 9, thickness profiles for flow over the microstructure: s, predicted;
experimental; s, theoretical. [, experimental.

coordinate system for the flow of a viscous liquid over a

macrostructure. There are slight discrepancies in rela-
tive thickness, but the overall shape of the free surface
is modeled accurately by eq 8.
Flow Patterns over the Microstructure of Or-
dered Packings. The amplitude and wavelength of
the microstructure of ordered packings are somewhat
smaller than the amplitude and wavelength of the
macrostructure. The characteristic Bond numbers in
microstructure are NBo ) O(10-1), and capillary forces
cannot be neglected. In principle, none of the terms in
eq 5 can be neglected a priori, and the complete equation
should be solved numerically.
For the sake of numerical convenience, in this analy-
sis the only dimensionless scale for all length variables
used was the length of the period of the microstructure.
Figure 4. Comparison of predicted and experimental free surface Details of the solution of this equation are given in
positions for flow over the macrostructure: s, predicted free Shetty and Cerro (1996). The film evolution equation
surface position; [, experimental free surface position.
is a third-order nonlinear ordinary differential equation
subject to periodic boundary conditions that was solved
experimentally- and theoretically-computed film thick- numerically by transforming it into a system of three
nesses is remarkably good, and the discrepancies be- first-order ordinary differential equations and using the
tween theory and experiments are within the limits of ISML subroutine BVPFD.
experimental error. Most striking is the coincidence of Figure 5 shows free surface profiles for flow over the
the positions of the maxima and minima of the experi- microstructure for 0.0209 < δ < 0.1088. These free
mental and theoretical film thicknesses. Since the surface profiles are qualitatively different from those
geometry of the solid surface is well-defined, the location obtained for flow over the macrostructure (Figure 2).
of the free surface is calculated simply as: Liquid fills up the concave sections of the solid surface
as the flow rate is increased. At sufficiently large values
yfs(x + h(x) sin θ) ) ys(x) + h(x) cos θ (8) of δ the gas-liquid interface becomes flat. Film thick-
ness profiles for flow over the microstructure are single
Note that the x-coordinate of the free surface location periodic in nature, as shown in Figure 6, where theo-
calculated from the normal film thickness at x is (x + retical and experimental normal free surface thick-
h(x) sin θ). Figure 4 is used to compare experimental nesses are compared for flow over a surface with a
and theoretical free surface positions in a laboratory periodic microstructure. Experimental and theoretical
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 775

Figure 7. Comparison between experimental and predicted free

surface profiles for flow over the microstructure: s, predicted; [,
experimental. Figure 9. Comparison between experimental and predicted film
thickness profiles for flow over the macrostructure kept with its
corrugations inclined: 9, experimental; s, predicted.

with the provision that the local angle of flow is equal

to the effective angle of flow, ω(x), with respect to the
vertical. The streamline function for this flow can be
written as:

[ ]
3
3q (y - ys)
Ψ(x,y) ) sin ω(x) h(x) (y - ys)2 -
3
2h (x) 3
(9)
The physical components of the gravity vector in
Cartesian coordinates acting on the normal and tan-
gential directions to the flow (i.e., X- and Y-directions
as defined in Figure 2) are

cos θ sin θ
gt ) g ; gn ) g (10)
sin ω sin ω
Under these definitions, the Nusselt film thickness
for a film flowing down a plane inclined at an angle ω
with respect to the vertical is given as:

(g 3νq
1/3

sin R)
h* ) (11)

The local film thickness is found by solving for the

Figure 8. Schematic of the macrostructure kept with its cor- zero-order perturbation of the film evolution equation,
rugations inclined. such as was done with eq 5 for flow over the macro-
structure:
free surface profiles on a laboratory coordinate system
are shown in Figure 7. sin ω(x)
Flow over the Macrostructure Where Corruga- φ0(x) ) (12)
[cos θ(x)]1/3
tions Are Inclined with Respect to the Vertical.
When the corrugations of the complex surface are at a Figure 9 is used to compare experimental and pre-
nonzero angle R with respect to the vertical, secondary dicted film thickness profiles for flow over the macro-
flows are generated by the relative motion of the fluid surface where the corrugations are at a 45° angle with
with respect to the direction of gravity forces as it flows respect to the vertical. These profiles are double-
over the corrugations (Figure 8). For plane-corrugated periodic when compared with the wavelength of the
surfaces Spekuljack (1986) demonstrated that the ef- periodic solid surface.
fective angle of flow of the fluid with respect to the The purpose of this introduction was to demonstrate
horizontal, ω(x), is a complex function of the solid the level of description attained with the theoretical
surface geometry, the corrugation angle of the packing, models of flow developed for complex surfaces. The
â, and the inclination angle of the corrugations with hydrodynamic descriptions will be used subsequently
respect to the vertical, R. The solid surfaces used in to evaluate important parameters for determining the
this work have cross sections of alternating half-circles performance of ordered packing surfaces.
(Figure 1); therefore, there is not a single corrugation
angle as in certain types of commercial packings. The 3. Computation of Liquid Holdup
macrostructure inclined at an angle other than 90o with
respect to the vertical can be analyzed using the same Underlying the models developed in this section is the
approach already used for the horizontal corrugations, assumption of uniform spatial distribution of liquid on
776 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

the entire packing surface. A perfect uniform liquid

distribution is an almost unachievable goal. To improve
the precision of the correlations presented here, it is
necessary to introduce assumptions about the degree
of maldistribution found in practical applications. The
degree of maldistribution is strongly dependent on the
intial distribution of liquid over the packing surface, and
it is very difficult to measure experimentally. The
model presented here, however, can be easily modified
to account for a given degree of maldistribution, if this
is known.
Liquid holdup, hL, is the volume of liquid present in
the void spaces of the packing, per unit volume of
packing. The bounds for the liquid holdup are 0 e hL
e , where  is the packing void fraction. Under flooding
conditions, hL = . Large values of liquid holdup
increase column pressure drop, the weight of the pack-
ings, the support load at the bottom of the packings,
Figure 10. Effect of liquid viscosity on liquid holdup. Kinematic
and column drainage time. Most importantly, when viscosity: s, 200 cSt; - ‚ -, 100 cSt; - -, 10 cSt; ‚‚‚, 1 cSt.
distilling thermally unstable materials, large holdup
may lead to excessive product degradation and fouling
(Kister, 1992).
Two types of liquid holdup are sometimes recog-
nized: static holdup and dynamic holdup. The static
holdup is the liquid remaining on the packing surface
after it has been fully wetted and drained for a long
time, i.e., holdup at zero liquid velocity, and it is a
function of liquid properties and packing geometry. Shi
and Mersmann (1985) presented an expression for the
thickness of the static holdup film on the packing
surface. Dynamic holdup is the liquid holdup during
flow and is, in addition to liquid properties and packing
geometry, a function of gas and liquid flow rates. The
dynamic liquid holdup or simply the liquid holdup, hL,
can be calculated by knowing the film thickness profiles
of the flowing film on the packing surface, assuming that
the entire packing surface is wet. A numerical integra-
Figure 11. Effect of macrostructure size on liquid holdup. Solid
tion of the film profile is used to calculate the volume surface period: s, 0.25 in.; - ‚ -, 0.5 in.; - -, 0.75 in.; ‚‚‚, 1.0 in.
of the liquid film per unit volume of the packing sur-
face. expression for the liquid holdup in dimensionless forms
is
4
hL ) 2
a
∫x)0 [yfs(x + hN(x) sin θ) -
x)a

hL ) 6.096NRe1/3NGa-1/3 (16)
ys(x + hN(x) sin θ)] dx (13)
Φ(X) cos(θ(X)) where NRe ) 4Fq/µ is the Reynolds number of the film
hL )
4h*
a
∫X)0
X)1

[sin ω]1/3
dX (14) flow and NGa ) ga3/ν2 is the Galileo number defined on
the basis of the characteristic length of the packing and
the fluid kinematic viscosity. Equation 16 has no
where hN(x) is the local normal film thickness in a
adjustable parameters. The constant C depends on
laboratory coordinate system and a is the period of the
packing geometry and can be easily found by numerical
solid surface. This definition of holdup is similar to the
integration, although its numerical values are very close
one given by Spekuljak (1986). The film thickness hN-
to the one reported here. This correlation must be taken
(x) can be calculated using the solutions to the film
as an upper bound for packings that are completely
evolution equations developed in previous sections of
wetted by the fluid and is valid for gas flow rates below
this paper. For flow over a surface with microstructure,
flooding.
i.e., δ > 0.2, the film thickness can be calculated using
Figure 10 shows the predicted liquid holdup on the
the solution of the complete film evolution equation,
macrostructure of a packing surface for a wide range of
including the capillary terms. For flow over the macro-
viscosities and liquid flow rates per unit area of packing.
structure, i.e., for cases where 0.1 < δ < 0.2, a simple
The void volume available, roughly  - hL, decreases
analytical expression derived from eq 12 can be used to
significantly even at moderate liquid flow rates for
compute hL. The integral shown in eq 14 is a number,
increasing liquid viscosity. As expected from eq 16,
C, computed on purely geometric terms.
liquid holdup increases as the 1/3 power of liquid mass
flow rate.
C) ∫X)0
X)1
(sin ω)2/3(cos θ)2/3 dX (15) Figure 11 shows the predicted liquid holdup as a
function of liquid flow rate for different packing sizes.
For a solid surface made of semicircular surfaces, These computations were done for a liquid of kinematic
a numerical integrationsthere is no closed-form viscosity 1 cSt. Liquid holdup increases significantly
integralsgives C ) 1.6775. Using this constant, the as the flow channels become narrower. The lines of
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 777

Figure 13. Effect of solid surface period on effective interfacial

area. Kinematic viscosity: O, 100 cSt; *, 10 cSt; 0, 1 cSt.

Figure 12. Comparison of liquid holdup between wetting and error. As a consequence, very few investigators report
nonwetting systems (this work). Wetting system: f, Billet (1989); actual interfacial area measurements with structured
9, Woerlee et al. (1994). packings of any kind. In order to estimate the interfa-
cial area, one must assume that the packing is com-
liquid holdup versus flow rate for different packing sizes pletely wet or that the wetted area is a known percent-
of the macrostructure are parallel to each other (period age of the area of dry packing.
) 0.5, 0.75, and 1.0 in.). Results for the smaller Free surface profiles obtained by integration of the
corrugation size (a ) 1/4 in.) however, show a qualita- film evolution equation can be numerically integrated,
tively different behavior, and the line is nearly hori- over the surface area of ordered packing, in order to
zontal. For small packing sizes eq 16 no longer applies calculate interfacial areas. The interfacial area per unit
because, for δ > 0.2, liquid film thickness is effected by width of the packing surface can be computed by the
capillary forces and holdup is a function of surface following equation:
tension.
Wetting characteristics of the liquid-solid systems
have a strong effect on liquid holdup. Figure 12 shows Aw ) 2 ∑xdyfs2 + dx2 (17)
a comparison between liquid holdups from two different
experimental sources (Billet, 1989; Woerlee et al., 1994). Equation 17 can be used to calculate either the solid
In both cases the liquids have approximately the same surface area (y ) ys) or the free surface area (y ) yfs)
kinematic viscosity, ν ≈ 1 cSt, but supposedly different per unit width. The effective area is defined as the free
wetting characteristics, for a packing with a similar surface area divided by the solid surface area per unit
specific area, 250 m2/m3. The liquid holdup predicted packing surface width. The interfacial area per unit
by eq 16, assuming a perfectly wetting film of liquid on column volume becomes
the packing surface, is always larger than the experi-
mentally observed liquid holdup for the air-water
system determined experimentally. Water-metal sys-
tems are notoriously nonwetting, and holdup would be
4∑xdyfs2 + dx2
ae ) (18)
lower than predicted for an ideal wetting system. The
values of liquid holdup for large liquid flow rates a2
approach the same values for both the wetting and the
nonwetting system when thin rivulets turn to wide (fat) Equation 18 assumes a smooth liquid surface without
rivulets and fat rivulets turn into films covering the waves and does not account for the increased interfacial
entire packing surface. Woerlee et al.’s (1994) data are area of waves caused by instabilities of the liquid film
given for metal-water-air also but appear to wet the or for the effect of shearing due to gas flow. Figure 13
surface of the packing more uniformly, even at low shows the effect of the solid surface period on the
liquid flow rates, although it is slightly shifted down, effective interfacial area. The effective area, defined as
indicating that not all the surface of the packing is ratio of free surface area to solid surface area, is close
wetted at all times. to 1 for the macrostructure where the liquid film closely
In the absence of reliable methods available in follows the contour of the solid surface. The area
literature to predict liquid holdup in structured pack- efficiency of a packing with small corrugations and/or
ings, eq 16 provides a simple but effective way of microstructure is smaller than the efficiency of a pack-
calculating the dynamic liquid holdup and can be used ing with large corrugations and no microstructure. The
to improve the methods available to compute flooding overall interfacial area per unit volume of the former
conditions in the presence of gas flow. packing, however, will be larger than the total interfa-
cial area of the larger-sized packing because more area
4. Interfacial Area of the smaller-sized packing can be fitted per column
volume.
In design practice the main objectives in packing While the computed efficient area for the packings
selection for a distillation unit are to minimize pressure used in this work never exceeds 1, there is experimental
drop and maximize specific interfacial area. Interfacial evidence (see de Brito et al. 1994; Fair and Bravo, 1990)
mass-transfer rates are directly proportional to the indicating that additional mass-transfer area may exist
effective gas-liquid interfacial area available during beyond the geometric nominal value due to instabilities
operation. Experimental methods to measure the actual on the liquid-gas interface (ripples, waves, detachment,
surface areas are complex and have large margins of and subsequent fragmentation of the film into droplets).
778 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

Figure 14. Interfacial area as a function of liquid flow rates: 9,

macrostructure; [, microstructure.
Figure 16. Predicted wetting efficiencies of packings from
literature holdup data: f, Billet (1989); 9, Woerlee et al. (1994).

Wetting phenomena are responsible for the formation

of dry patches and result in uneven local distribution
of liquid. These phenomena, related to static and
dynamic contact angles and to the nature of the three-
phase contact line, add a completely new dimension to
the hydrodynamics of ordered packings and are not
considered in this publication. One may use, however,
the information on liquid holdup and our models to
estimate the percentage of packing surface that is not
covered by a liquid film. Assuming no edge effects in
the liquid rivulets, for a given overall flow rate the ratio
of the wet area to the total area is given by a simple
relationship

( )
(hL)exp 3/2
Figure 15. Schematic of free surface profiles of liquid film flowing η) (19)
over the macrostructure with the microstructure superimposed on (hL)calc
it.
where (hL)exp is experimental holdup and (hL)calc is
Figure 14 shows the interfacial areas calculated using computed holdup for a given overall flow rate.
eq 17 for the macrostructure and the microstructure of Using eq 19, it is possible to use available experimen-
packing as a function of the liquid load. The interfacial tal data to compute the corresponding wet area of
area due to the film flowing over the macrostructure is packing. Figure 16 shows the percentage of wet area
relatively constant. For a packing surface with a computed using eq 19 for the experimental data of Billet
nominal interfacial area of 250 m2/m3, this is in good (1989) and Woerlee et al. (1994). Notice that at small
agreement with experimental results of de Brito et al. liquid flow rates Billet’s (1989) data show that the liquid
(1994). The presence of the microstructure increases is wetting only a very small fraction of the packing
the interfacial area available even when the increased surface. This fraction increases promptly with increas-
area drops off rapidly when liquid flow rate increases. ing flow rate to reach about 90% of the total area.
This effect is shown schematically in Figure 15 where Woerlee et al.’s data (1994) indicates a larger fraction
a microstructure and a macrostructure coexist. At low of wetted packing, even at small flow rates. This may
flow rates the liquid flows as a thin film over the be due to special surface treatment used in these
microstructure. The microstructure provides an added packings in order to decrease static contact angles. The
amount of interfacial area available for mass transfer fact that the percentage of wet area never increases past
between phases. As the flow rate is increased, the 80% may be due to liquid maldistribution on top of the
“valleys” of the microstructure quickly fill up, as was packing.
shown in Figure 5. The flow reaches what we call the
“microstructure” limit, and from then on the packing 5. Contact Time
surface may behave as a smooth macrostructure would.
At very high flow rates, even the macrostructure fills A liquid film flowing by action of gravity over a
up and flooding occurs. The microstructures considered complex surface follows the potential energy gradient
here are only 1/4 of the size of corrugations of the along the surface. For a packing surface with plane
macrostructure. In industrial practice the characteristic corrugations the potential energy gradient on the flat
size of the microstructure is less than an order of sections of the surface is oriented at an angle that can
magnitude smaller than the characteristic size of the be calculated based on a relatively simple geometrical
macrostructure. Therefore, while the additional inter- construction involving the corrugation angle, â, and the
facial area created by the microstructure is large, the inclination angle, R (Spekuljak, 1986). For the rounded,
microstructure flow rate limit will be attained at even half-circle packing surfaces used in this work, the
lower flow rates (dotted lines in Figure 14). Microstruc- potential energy gradient depends continuously on the
tures, however, have other purposes such as preventing local angle θ (see Figure 8). The analytical construction
film breakup and promoting wave formation. of the gradient of a complex surface on a three-
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 779
flowing inside the valley, the velocity of the free surface
is computed using a flow angle equal to the inclination
angle of the corrugations, R, for a half-circle surface or
the inclination angle defined by Spekuljak (1986) for a
corrugated surface. During the rest of the path, for all
types of surfaces, the liquid follows a vertical path line
as it goes over a peak of the corrugation.
Computation of the length of the path that the liquid
follows down the solid surface can be done by recogniz-
ing two contributions that can be compounded to find
the total length. First, a fluid particle moving on a
surface where corrugations are inclined by an angle R
is moving along a path that has been “stretched” out
by a factor of 1/sin R. This simple relationship is used
to compute the vertical distance between two peaks, Linc.
The second contribution is due to the fact that particles
move, not on a vertical path, but at an angle determined
by the gradient of the potential energy, as was already
discussed in this section. The actual length of the path
traveled by a fluid element between two peaks is larger
and can be computed by taking into account the
stretched length when the particle flows over the peaks
and the length that particles follow when they are inside
Figure 17. Path followed by liquid flowing over the macrostruc-
ture.
the valleys. The first effect mentioned here, that is,
increased length due to the inclination of the surface,
dimensional space is a possible but formidable task. To is reduced by the same amount when the overall height
determine contact times for all types of packing surfaces, of packing is measured. The second effect, however, is
a simplified method was used, based on flow visualiza- more complex because, by changing the inclination
tion experiments designed to map path lines of tracer angle R, the length of the path of the particles in the
particles moving within the liquid film. A description valleys is increased but the average velocity is also
of these experiments follows. increased.
After steady flow of 5 cSt silicone oil had been Taking into account these contributions, the contact
established on a test surface, a very small drop of a time for one surface cycle can be computed using the
colored food dye, that had been intimately mixed with following relationship:

( )
the same silicone oil, is introduced on top of the test
surface. A CCD camera with a Nikkor 55 mm lens and h(x) x(xs1 - xs2)2 + (ys1 - ys2)2
a PK-13 extension ring was used to record the path of
the tracer particle. The images were stored on tape
τ) ∫x)0
x)a

1.5q sin ω sin R sin(90 - θ)

dx
using a Panasonic AG7300 video recorder for later (20)
analysis. A magnification of up to 4× was obtained The denominator of the integral in eq 19 represents
using this setup. The images were later played back the film surface velocity, while the numerator repre-
using a frame shuttle to control the frame speed, and a sents the arc length distance measured along the
tracing of the path line that the tracer particle followed packing surface.
was made from the Sony PVM-2530 display monitor. For a packing surface with corrugations inclined at
Figure 17 shows schematically such path lines for two an angle R, the x-axis is stretched out by a factor 1/sin
different flow rates. The diagonal solid lines represent R. This arc length integrated from one peak to another,
the peaks of the solid surface. For the entire range of and, consequently, the ratio of the actual length of the
flow rates covered in this experiment, path lines fall path to the length of the path for R ) 90°, is a function
exactly on the same place; i.e., the path lines are not a only of R. This ratio, designated by lratio, goes from lratio
function of flow rate and, being free of inertial effects, ) 1 when R ) 90° to lratio f ∞ when R ) 0. For the
follow the orientation of the potential energy gradient. range of inclination angles of practical interest, 25° <
For the test surface made of half-circles, except at the R < 65°, this ratio can be computed using the following
peaks the liquid follows closely the angle of inclination correlation:
of the corrugations. The direction of the potential
energy gradient changes very littleswithin 5°salong
lratio ) 3.7617 - 0.12299R + 0.001976R2 -
the half-circle that forms the valley of these surfaces
and changes abruptly to almost 90° when it flows over 1.1167 × 10-5R3 (21)
the half-circle that forms the peak. When flowing over
a corrugated surface with almost a triangular cross There are no adjustable constants in eq 21 outside
section, liquid follows the potential energy gradient on the approximation introduced by the correlation given
the straight section of the corrugation and then changes in eq 20. Using the definition of lratio, eq 20 reduces to:

( )
abruptly to 90° to go over the bend to the next flat
section. τν 7.268lratio
) Nret ) NRe-2/3NGa-1/3 (22)
Contact time for the liquid flowing down on an a2 sin R
inclined packing surface is computed using the analyti-
cal expression for the velocity of the free surface (eq 2). where Nret, the retention number, is a function of the
Due to the periodic nature of the flow, contact time can lratio, the inclination angle, and the Reynolds and Galileo
be estimated by integration of eq 2 along a path line numbers. The constant C ) 7.268 is the result of
between two peaks. During part of the path, while integrating the geometrical parameters of the packing
780 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

and, as the one given in eq 15, changes very little with

packing geometry.

C) ∫X)0
X)1 1
(cos θ)4/3
dX (23)

In short, contact times depend on the inclination angle

of the packing and can be considerably larger than
contact times computed under the assumption that the
flow of the liquid down the packing surface is analogous
to film flow down the vertical walls of a wetted wall
column (de Haan and de Graauw, 1991; Rocha et al.,
1993).

6. Liquid-Side Mass-Transfer Coefficient

Prediction of mass-transfer rates is one of the major
Figure 18. Comparison of liquid phase mass-transfer coefficient
difficulties in the design of packed columns. Few with empirical correlation of Bravo et al. (1990): s, this work; ‚‚‚,
researchers have attempted to predict mass-transfer Bravo et al. (1990).
rates on the basis of the fluid flow patterns inside
packed beds. The first attempts to develop a mecha- needed to accomplish a standard separation. When the
nistic-theoretical model for structured packings were equilibrium and operating lines are straight and parallel
done by Bravo et al. (1985) and Spekuljak and Billet to each other, the number of transfer units required for
(1987), based on a wetted-wall analogy. Their models a given separation is equal to the number of theoretical
can be used as a reasonable basis for predicting mass- plates or stages required to carry out the same separa-
transfer rates between bulk vapor and liquid phases. tion in a stagewise mass-transfer device. In general,
The main assumption, that the packing surface was the number of transfer units and the number of theo-
totally wetted, allows the interfacial area to be equal retical stages are a measure of the difficulty of the
to the specific surface area of the packing. There are separation. The heights of the transfer units for the
three basic approaches to predict liquid-side mass- individual phases can be written as:
transfer coefficients, i.e., (i) film theory, (ii) penetration
theory, or (iii) surface renewal theory. Since liquid flows Ug Ul
over short inclined planes before being remixed in each HTUV ) ; HTUL ) (26)
kgae klae
sheet fold, the film theory of Whitman (Coulson and
Richardson, 1970) is inadequate. For short contact Here Ug and Ul are the gas and liquid superficial
times, the model which best fits experiments is Higbie’s velocities based on the empty column diameter and ae
penetration theory (Treybal, 1968). For laminar flow is the interfacial area per unit column length. The
of a thin liquid film, we write overall height of a transfer unit can be written using
the gas and liquid heights weighted by the ratio of gas

x
Dl and liquid mass flow rates:
k1 ) 2 (24)
πτ HTUOG ) HTUV + λHTUL (27)
In this expression Dl is the liquid diffusion coefficient
and τ is the contact time for one cycle of the ordered where λ ) mG′/L′. G′ and L′ are the gas and liquid
packing. Substitution of the expression for contact molar flow rates, and m is the slope of the equilibrium
times developed in section 5 (eq 22) renders line. Another quantity that can be used to measure the
efficiency of packings for carrying out separations is the
height equivalent to a theoretical plate, HETP. The
x
sin R
NSh ) 0.4185 N 1/3NGa1/6NSc1/2 (25) relationship between HTU’s and HETP’s has been
lratio Re shown to be (Sherwood et al., 1975)
Figure 18 compares values of kl predicted using eq ln λ
25 and values computed using the widely used empirical HETP ) HTUOG (28)
λ-1
correlation proposed by Fair and Bravo (1990) for
absorption of CO2 in an air/water system. Values Theoretical values of HTU and HETP can be calcu-
predicted using eq 24 are for flow over the “half-circle” lated using velocity profiles, interfacial areas, contact
macrostructure of a 1 in. period. The values computed times, and liquid-side mass-transfer coefficients pre-
using the empirical correlation are for flow over the dicted using the methods developed in this work.
Montz-B1 which is also a rounded packing having a Figure 18 is used to compare computed values of HTU
specific area (250 m2/m3) similar to that of the theoreti- with data of Spekuljak and Billet (1987) for the test case
cal macrostructure. These kl values are in good agree- of CO2 desorption from water by means of air counter-
ment, validating the more fundamental approach used current flow at atmospheric pressure and ambient
in this work. Notice that eq 25 does not have any temperature. Spekuljak (1986) used a 0.225 m i.d.
arbitrary, adjustable constants. column, 1 m in height filled with a Montz-B1 sheet
packing of specific surface area 325 m2/m3. The gas flow
7. HTU and HETP rate was 120 m3/h. The composition of CO2 varied
between 0.22 and 0.0107 mole fraction. All data were
The height of the transfer unit, HTU (Chilton and obtained at a pressure of 760 mmHg and between 18
Colburn, 1935), is a measure of the height of packing and 20 °C. Physical properties used in these computa-
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 781

Figure 20. Comparison of predicted HETP values with experi-

mental from FRI and literature model of Bravo et al. (1992) at
Figure 19. Comparison of predicted HTU with literature 100 psia: f, FRI (experimental); s, predicted (this works100%
correlations: s, this work (100% wetting); - ‚ -, this work (75% wetting); ‚‚‚, predicted (this works75% wetting); - - (this
wetting); ‚‚‚, this work (50% wetting); 9, Spekuljaksexperimental works50% wetting); - ‚ -, Bravo and Fair (1992).
(1986); *, Spekuljakspredicted (1986); 0, Zogg (1972).

tions were calculated at 1 atm and 19 °C. A rounded

model approximation of the Montz-B1 type packing was
used to compute the parameters for the theoretical
HTU’s. The model has a period a ) 0.0193 m and
amplitude R ) 0.00483 with an amplitude to period
ratio of 0.25 and a specific surface area of 325 m2/m3.
The CO2/water system has been extensively studied for
packed columns, and it is safe to assume that mass
transfer is controlled by the liquid-phase resistance and
that the height of the transfer unit is independent of
gas flow rates for Ul/Ug < 1 (Perry and Chilton, 1973).
Under these assumptions HTUV can be neglected in the
computation of the HTU’s.
There is no good agreement of our model with Spe-
kuljak’s data for small liquid flow rates. At large liquid Figure 21. Comparison of predicted HETP values with experi-
flow rates the agreement is better but still off by about mental from FRI and literature model of Bravo et al. (1992) at
20%. In this type of experimental system it will be 300 psia: f, FRI (experimental); s, predicted (this works100%
almost impossible to have complete wetting or to assume wetting); ‚‚‚, predicted (this works75% wetting); - - (this
that all packing areas are working at 100% capacity. works50% wetting); - ‚ -, Bravo and Fair (1992).
Assuming that only a fraction of the packing is wet, one
can compute an effective flow rate as well as effective ficients were computed using Bravo et al. (1992) cor-
liquid holdup, contact times, and liquid-side mass- relation:

[ ]( )
transfer coefficients. Under these assumptions it is
kGa 0.8 0.33
straightforward to compute values of HTU for only a (UGe + ULe)FGS µG
partially wetted packings. These results are shown in NShG ) ) 0.054
DG S µG DGFG
Figure 19 as dotted lines for 50% and 75% wetting (29)
effectiveness. Spekuljak’s experimental data could be
interpreted easily if one allows for only partial wetting where UGe and ULe are the effective liquid and gas
of the packing surfaces. velocities. HETP values were computed using eqs 25-
Correlations developed in this paper were also com- 29. Figures 20 and 21 compare experimental data from
pared with data provided by Fractionation Research Inc. FRI at 700 and 2100 kPa with HETP predicted using
of Stillwater, OK. FRI data were obtained using a the theoretical, no-adjustable parameters model devel-
Mellapack 250Y sheet metal packing in a true com- oped in this work as well as the empirical model from
mercial scale column, 1.2 m in diameter. There were Bravo et al. (1992). Values of HETP are plotted against
18 layers of packing, each one 0.21 m in height for a Fs, a measure of the gas kinetic energy. Agreement with
total height of 3.78 m. Each layer was rotated 90° with experimental data is good at low pressures and an
respect to the layer below. These tests were conducted intermediate range of flow rates. At low gas and liquid
with the isobutane/n-butane system at 700, 1140, 2100, flow rates one would expect partial wetting, and at high
and 2700 kPa (100, 165, 300, and 400 psia, respectively). liquid gas and liquid flow rates, partial flooding may
Liquid flow rates varied from 6.5 to 103.0 m3/h (28.5 to be present. The dotted lines represent 75% and 50%
450 gpm), and gas flow rates ranged from 3409 to 45455 partial wetting. Except for the very small values of gas
kg/h (7.5-100 mLb/h). Experimental HETP values and liquid flow rates, most experimental data fall within
were calculated from data taken at total reflux condi- these bounds. At the higher pressure shown in Figure
tions. 21, agreement is not as good for any of the two
Predicted liquid-side mass-transfer coefficients were correlations shown. Film thickness and contact times
computed using eq 28. Gas-side mass-transfer coef- were computed using actual values of liquid viscosity
782 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

and diffusivity. At the higher pressures (and temper- g: acceleration due to gravity, m/s2
atures) of these tests, liquid viscosities are nearly half hL: liquid holdup, m3/m3
the viscosity at low pressure while liquid diffusivities h(x): film thickness, m
are more than twice that at low pressure. For the same h*: Nusselt film thickness, m
liquid flow rates, at high pressure liquid film thickness H*: dimensionless Nusselt film thickness
would be smaller, due to the reduced viscosity, and H: mean curvature of free surface, 1/m
penetration would be much larger, due to the increased H: dimensionless mean curvature
diffusivity. As a consequence, the assumptions neces- HETP: height equivalent of theoretical plate, m
sary for penetration theory to hold may not be valid. HTU: height of transfer unit, m
When experiments are done at total reflux and the same k: mass-transfer coefficient, m/s
volumetric gas flow rate is maintained, this statement l: distance between peaks of solid surface, m
would not be true. There is also the question of in- lratio: ratio of actual path length to path for R ) 90°
creased viscosities and densities of the vapor phase with L: liquid mass velocity, kg/s
the corresponding increase in entrainment. In struc- m: slope of the vapor-liquid equilibrium line
tured packings where lateral movement is low, it is NBo: Bond number, FgA2/δ
more difficult for the vapor to disengage from the liquid. NCa: Capillary number, µU/σ
In addition, low surface tension may increase foaming. NGa: Galileo number, ga3/ν2
All these factors contribute to generate actual operating NRe: Reynolds number, 4Fq/µ
conditions different from the flow model presented here, NShe: Sherwood number, kLa/DL
with the corresponding decrease in efficiency. NSc: Schmidt number, ν/D
q: flow rate/unit width, m3/s‚m
8. Concluding Remarks Q: flow rate/unit width, m3/h‚m.
R: radius of “half-circle” surface, in.
The foundations for a fundamental approach to the
Ul: liquid superficial velocity, m/s
design of ordered packings have been set using basic
Ug: gas superficial velocity, m/s
fluid mechanics and mass-transfer results. This ap-
V: velocity, m/s
proach allows prediction of liquid holdup, interfacial
x, X: dimensional/dimensionless main coordinate direction
areas, mass-transfer coefficients, and the height of
y: coordinate direction normal to flow
mass-transfer units without using a single adjustable Ys: solid surface coordinate
parameter. Predictions of design parameters using the
methods proposed in this paper are in good agreement Greek Letters
with experimentally observed values and trends. R: angle of corrugations with respect to vertical, rad
More work needs to be done to improve predictions â: angle of inclination of packing corrugations with respect
and to understand other phenomena on a rigorous, to vertical
fundamental basis. One obvious area of improvement δ: h*/A
will be to use velocity profiles, film thickness values and : packing void fraction
free surface positions that can be predicted using the φ: dimensionless film thickness
film evolution equations developed in this paper and to θ: azimuthal location, deg
solve the mass-balance equations for the liquid film λ: ratio of slopes, equilibrium line/operating line
flowing down the packing surface on a volume-conserved η: wetting efficiency
basis. This approach will bypass the use of approximate F: density, kg/m3
mass-transfer theories to compute mass-transfer rates. µ: viscosity, kg/m‚s
Another area of work is clearly related to wetting and ν: kinematic viscosity, m2/s
spatial liquid distribution inside the packings. This has τ: contact time, s
been recognized as one of the weak areas of knowledge σ: surface tension coefficient, N/m
and is much wanting on experimental as well as ω: effective angle seen by fluid flowing over inclined
theoretical work. corrugations
The main advantage of the basic approach used in ψ: streamline function
ø: film wavelength number, A/a
this paper is that it is possible to analyze the perfor-
ú: perturbation parameter, δø
mance of distillation and absorption units by identifying
what are the assumptions of this model that are not Subscripts
completely fulfilled or that can be improved by further c: coordinates of the center of the half-circle
modeling. fs: free surface
g: gas phase
Acknowledgment inc: inclined
The authors acknowledge the National Science Foun- l, L: liquid phase
dation for their research Grant CTS-8912784, which m: microstructure
allowed this research to get started. Support from the N: normal to the solid surface
Office of Research of The University of Tulsa is greatly O: overall
acknowledged. The authors thank Dr. John Kunesh s: solid surface
and Fractionation Research Inc. for access to some of static: static holdup, i.e., at zero liquid velocity
the data used in this paper. V: vapor

Nomenclature Literature Cited

A: amplitude of solid surface, m
Billet, R. Packed Column Analysis and Design; Ruhr University:
a: solid surface period, m
Bochum, Germany, 1989.
ae: interfacial area, m2/m3 Bravo, J. L.; Rocha, J. A.; Fair, J. R. Mass Transfer in Gauze
Aw: interfacial area/unit packing width, m2/m Packing. Hydrocarbon Process. 1985, 64, 91.
C: constants defined in eqs 15 and 23. Bravo, J. L.; Rocha, J. A.; Fair, J. R. Pressure Drop in Structured
D: diffusion coefficient, m2/s Packings. Hydrocarbon Process. 1986, 56, 45.
 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 783
Bravo, J. L.; Rocha, J. A.; Fair, J. R. A Comprehensive Model for Shetty, S. A. Liquid Spreading and Thin Film Flows Over Complex
the Performance of Columns Containing Structured Packings. Surfaces. Ph.D. Dissertation, The University of Tulsa, Tulsa,
ICHEME Symp. Ser. 1992, No. 128, A489-A507. OK, 1995.
Chilton, T. H.; Colburn, A. P. Distillation and Absorption in Packed Shetty, S. A.; Cerro, R. L. Flow of a Thin Film Over a Periodic
Columns: A Convenient Design and Correlation Method. Ind. Surface. Int. J. Multiphase Flow 1993, 19 (No. 6), 1013.
Eng. Chem. 1935, 27, 255. Shetty, S. A.; Cerro, R. L. Liquid Film Flow over the Microstruc-
Coulson, J. M.; Richardson, J. F. Chemical Engineering; Pergamon ture of an Ordered Packing. Int. J. Multiphase Flow 1996,
Press: Oxford, U.K., 1970. submitted for publication.
de Brito, M. H.; von Stockar, U.; Bangerter, A. M.; Bomio, P.; Laso, Shi, M. G.; Mersmann, A. Effective Interfacial Area in Packed
M. Effective Mass-Transfer Area in a Pilot Plant Column Columns. Ger. Chem. Eng. 1995, 8, 87.
Equipped with Structured Packings and with Ceramic Rings. Spekuljak, Z. Modelacion de Rellenos Regullares de Alta Eficienca
Ind. Eng. Chem. Res. 1994, 33, 647. Para la Transferencia de Materia. Ph.D. Dissertation, Univer-
de Haan, A, B.; de Graauw, J. Mass Transfer in Supercritical sidad Nacional del Litoral, Santa Fe, Argentina, 1986.
Extraction Columns with Structured packings for Hydrocarbon Spekuljak, Z.; Billet, R. Mass Transfer in Regular Packing. Lat.
Processing. Ind. Eng. Chem. Res. 1991, 30, 2463. Am. J. Heat Mass Transfer 1987, 11, 63.
Fair, J. R.; Bravo, J. L. Distillation Columns Containing Struc- Stasiek, J.; Collins, M. W.; Chew, P. E. Investigation of Flow and
tured Packing. Chem. Eng. Prog. 1990, 86 (1), 19. Heat Transfer in Corrugated Passages. Experimental results.
Gaiser, G.; Kottke, V. Flow Phenomena and Local Heat and Mass Int. J. Heat Mass Transfer 1996, 39, 149.
Transfer in Corrugated Passages. Chem. Eng. Technol. 1989, Stichlmair, J.; Bravo, J. L.; Fair, J. R. General Model for Prediction
12, 400. of Pressure Drop and Capacity of Countercurrent Gas/Liquid
Hanley, B.; Dunbobbin, B.; Bennett, D. A Unified Model for Packed Columns. Gas Sep. Purif. 1989, 3, 19.
Countercurrent Vapor/Liquid Packed Columns. 2. Equations for Treybal, R. Mass Transfer Operations; McGraw Hill: New York,
the Mass Transfer Coefficients, Mass Transfer Area, the HETP, 1968.
and the Dynamic Liquid Holdup. Ind. Eng. Chem. Res. 1994, Woerlee, G. F.; Olujic, Z.; Graaw, J. An analytical approach to the
33, 1222. modelling of packed column hydraulic. AIChE Annual Meeting,
Kister, H. Z. Distillation Design; McGraw Hill, Inc.: New York, San Francisco, Nov 1994.
1992. Zhao, L.; Cerro, R. L. Experimental Characterization of Viscous
Kister, H. Z.; Gill, D. R. Predict Flood Point and Pressure Drop Films over Complex Surfaces. Int. J. Multiphase Flow 1992,
for Modern Random [Link]. Eng. Prog. 1991, 87, 32. 18, 495.
Kurtz, D. P.; McNulu, K. J.; Morgan, R. D. Stretch the Capacity Zogg, M. Stromungs-und Stoffaustauchuntersuchungen an der
of High-Pressure Distillation Columns, Chem. Eng. Prog. 1991, Sulzer Gewebepackung. Dissertation No. 4886, Eidgenossichen
87, 43. Technische Hochschule, Zurich, 1972.
Nawrocki, P. A.; Xu, Z. P.; Chuang, K. T. Mass Transfer in
Structured Corrugated Packing. Can. J. Chem. Eng. 991, 69,
1336. Received for review October 7, 1996
Perry, R. H.; Chilton, C. H. Chemical Engineers Handbook, 5th Revised manuscript received December 20, 1996
ed.; McGraw Hill: New York, 1973. Accepted December 22, 1996X
Rocha, A. J.; Bravo, J. L.; Fair, J. R. Distillation Columns
Containing Structured Packings: A Comprehensive Model for IE960627J
Their Performance. 1. Hydraulic Models. Ind. Eng. Chem. Res.
1993, 32, 641.
Sherwood, T. K.; Pigford, R. L.; Wilke, C. R. Mass Transfer; X Abstract published in Advance ACS Abstracts, February
McGraw Hill, New York, 1975. 1, 1997.